New diagnostic criteria for diabetes mellitus are (1) symptoms of diabetes mellitus plus a random plasma glucose concentration of at least 200 mg per dL (11.1 mmol per L); (2) a fasting plasma glucose level of 126 mg per dL (7.0 mmol per L) or higher; or (3) a two-hour plasma glucose level of 200 mg per dL or more during an oral glucose tolerance test. However, the oral glucose tolerance test is not recommended as a test for diabetes mellitus. Repeat testing on a different day is needed to confirm these findings.

Davidson and colleagues conducted a cross-sectional analysis to determine if, in fact, patients diagnosed with diabetes mellitus by these criteria had excessive glycosylation as defined by an elevated glycosylated hemoglobin A_1c (HbA_1c) level. Their rationale was that patients with normal HbA_1c levels do not have excessive glycosylation and should not be diagnosed with diabetes mellitus regardless of whether or not they meet the criteria for diabetes mellitus as listed above. They arrived at this conclusion because it is the excessive glycosylation that is thought to lead to the microvascular complications of diabetes mellitus, so “artificially” diagnosing diabetes mellitus when the patient does not actually have a condition that would lead to diabetic complications is unwise.

Data from the National Health and Nutrition Examination Survey (NHANES III) and from the Meta-Analysis Research Group (MRG) on Diagnosis of Diabetes were used in the study. Patients included from the NHANES III data were those between 40 and 74 years of age without a prior history of diabetes mellitus. In both groups, those analyzed had had appropriate blood levels of glucose and HbA_1c drawn. Normal HbA_1c levels were those below the mean plus two standard deviations. The “newly normal” group consisted of patients whose oral glucose tolerance test would have been abnormal by the old diagnostic criteria, but who did not have an abnormal fasting plasma glucose level. These patients would now not be considered diabetic. The “newly diabetic” group were patients whose fasting plasma glucose levels had previously been between 126 and 139 mg per dL (7.0 to 7.7 mmol per L), thereby falling out of the range for diagnosis of diabetes mellitus according to the old criteria. The goal of this study was to determine if glycosylation was higher in the newly diabetic group than in the newly normal group.

Of the 1,846 patients (out of 2,836 people eligible) in the NHANES III group with a normal glucose tolerance, the normal range for HbA_1c was 4.4 to 6.1 percent. The normal range in the MRG group was found to be 3.9 to 6.3 percent in the 6,952 patients with a normal glucose tolerance (out of 8,917 eligible patients). Less than 3 percent of those with normal fasting plasma glucose levels had excessive glycosylation. The prevalence of an elevated HbA_1c level was determined to be about 3 percent in those who were classified as newly diabetic. The prevalence of elevated HbA_1c levels was 49 percent in those whose fasting plasma glucose level was more than 139 mg per dL (7.7 mmol per L). Most of those (73 percent) who were newly diagnosed with diabetes mellitus had normal HbA_1c levels. None of these patients had HbA_1c levels above 7.1 percent. In the newly normal group, only 0.3 percent of patients had HbA_1c levels above 7.1 percent, with 18 percent having mild elevations. Overall, 60 percent of those determined to be diabetic according to the new criteria had normal HbA_1c levels.

The authors conclude that using the new criteria for diagnosing diabetes mellitus may lead to more harm than good, as more patients are diagnosed with diabetes mellitus. They argue that an HbA_1c level of less than 7.1 percent will not trigger any therapeutic changes, and that these patients (HbA_1c level less than 7.1 percent and fasting plasma glucose level of 126 to 139 mg per dL) should be classified as having impaired fasting glucose. This would eliminate what they call “false positive” diagnoses of diabetes mellitus.

In an accompanying editorial, Vinicor challenges the above study. The assumption that tissue glycosylation occurs in tandem with hemoglobin glycosylation may be faulty. Vinicor proposes that end-organs may be damaged at HbA_1c levels lower than those considered “abnormal.” Also, although Davidson and associates are essentially attempting to improve the specificity of tests to diagnose diabetes mellitus, their proposed method of determining a diagnosis of diabetes mellitus or impaired fasting glucose level requires multiple diagnostic tests. In addition, the definitions of elevated HbA_1c are, admittedly, arbitrary, and are not, in and of themselves, known to be associated with the occurrence of microvascular damage. Vinicor concludes that the current recommendations of the American Diabetes Association and the World Health Organization (as listed in the first paragraph above) are reasonable. Failure to identify a patient with diabetes mellitus is more serious than over-diagnosing this condition.